Media-spanning artist

1

Amsterdam Business School

Wouter Adriaan Jongejan 5970288

Master Thesis Topic: Media-spanning artist Program: MSc Business Administration

Specialization: Entrepreneurship and Management in the Creative Industries (EMCI) Supervisor: Monika Kackovic (Amsterdam Business School)

Academic Year: 2014/2015 Semester: 2nd

Statement of originality

This document is written by Student Wouter Jongejan (5970288) who declares to take full

responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

2 Abstract

Recent academic research towards the potential consequences of category-spanning have their theoretical roots in organizational theory. Lately, however, academics from within the creative industries also put their emphasis on this problem. Category-spanning is found to have potential negative effects on audience appeal and thus evaluation, but is on the other hand used as strategy to attract larger audiences and to differentiate from the crowd. While findings in literature are contradictive, this thesis is led by the following research question: “To what extent do differences in both media-spanning and artistic status change expert and non-expert art observers’ evaluations in terms of their perceived financial market value and aesthetic experience of paintings?”

Method

Via an online survey that has been answered by 231 respondents, collected data is analyzed with SPSS (correlation, regression and non-parametric t-tests). The respondents were assigned to seven different research conditions in which they all had to evaluate the same portrait by Pablo Picasso (different spanning and status cues). A median split is used to divide participants in 2 groups based on art expertize.

Results

Non-significant results indicate that an artist who spans different artistic media into its

communicated oeuvre, suffers in terms of perceived financial value compared to an artists who only works in a single medium (painting). Aesthetic experience of the media-spanner showed higher levels for the spanning artist. The differences based on status were even less significant. It has been argued that status cues were not strong enough to provoke any effect. Furthermore, none of the stated research hypotheses could hold and are therefore rejected.

Conclusion

It has been concluded that spanning does negatively influences the financial perception of an artwork in the eyes of different beholders. Although, the effects were not statistically significant, the implications for the creative industries cannot be unmentioned.

3

Table of Content

Role of status as quality cue for the appreciation of art 14

Aesthetic experience 16

The role of audience type on art evaluations: art expertise 17

Perceived value in monetary terms 19

Methodology 21 Research design 21 Respondents 22 Demographics 23 Method 24 Experimental treatments 25 Measures 27

Analyses and results 30

Descriptive statistics 30

Transforming data: - Financial value 33

- Aesthetic experience 34

Other variables 36

Correlations 37

Testing hypotheses with non-parametric t-tests 40

Perceived creativity 45

Regression analysis 47

- Spanning and status on perceived financial value 47

- Spanning and status on aesthetic experience 48

Discussion 50

General discussion 50

Limitations and suggestions for future research 57

Conclusion 59 References 60 Appendix 64 Research conditions Homogeneity of variances Normality distributions

4

Introduction

“Pablo Picasso(Spanish, 1881–1973), one of the most prominent, innovative artists of the 20th century, is celebrated for his lengthy and prolific career working in several modernist idioms, as well as for co-founding Cubism. Picasso was born in Málaga, Spain, and began drawing and painting early” (Artnet.com).

Pablo Picasso is, as we can learn from the quote above a successful example of an artist spanning different artistic media. Academic researches towards these category-spanning activities have their theoretical roots in organizational theory. Decades ago, researchers started to explore the effect of companies spanning their original organizational boundaries into new markets or categories (Zuckerman, 1999). Lately however, the applicability of these theories also went to other kinds of research. Hsu et al. for example conducted research on the niche width of movies spanning film genres (creative industries) and how this has effect on overall audience appeal (2012). The underlying premise is that what they call generalist (spanning multiple categories at the same time), are less efficient than specialist at exploiting business opportunities. That is because they spread their capabilities across multiple positions instead of focusing on a single position. The trade-off assumption for which she found empirical evidence in an earlier research is called the principle of allocation and suggests that “the more diversity in regions of resource space targeted by an organization, the lower the organization’s capacity to perform well within them” (Hsu, 2006 P. 420). Translated to the arts industry, it thus can mean that category-spanning artists also face problems concerning audience appeal, and furthermore influence observers’

perceptions of financial value and for example perceived aesthetic experience of certain works of art when engaging in spanning. However, as exposed by the example of Pablo

5 Picasso above, it could on the other hand also be a very viable act for certain artists. Other researchers like Sgourev and Althuizen (2014) found that artists with high status (e.g. Picasso due its prolific career) can benefit from what they call ‘stylistic inconsistency’ to their former works of art (earlier works in their oeuvre). This implies that they are working, and are capable of working in different artistic styles. Audience members do favor this act of inconsistency as being creative and won’t penalize the artists for such behavior and actually praise spanning compared to what was expected from most former organizational theory (people will penalize companies who do not stick to their original boundaries). However, as already mentioned, only high status artists did gain this advantage compared to low - and mediocre status artists as found by their study (Sgourev & Althuizen, 2014).

Sgourev & Althuizen furthermore used sets of consistent and inconsistent artworks in order to test their hypotheses on the research concepts: perceived financial value, market value and aesthetic evaluation as perceived by the participants based on stylistic spanning

activities (2014). This thesis is building upon their research, but is motivated by the following different question:

“To what extent do differences in both media-spanning and artistic status change expert and non-expert art observers’ evaluations in terms of their perceived financial market value and aesthetic experience of paintings?”

In order to receive an answer on this question, a questionnaire (N = 231) was conducted testing whether there will be differences between two participant groups (consumers) that are divided based on their level of art expertise (expert vs non-expert) and furthermore how they react towards the artistic act of spanning (in terms of artistic media used) in an

6 research conditions (1: control; 2: no spanning – high status; 3: no spanning: low status; 4: Spanning – high status; 5: spanning – low status; 6: no spanning – no status & 7: spanning – no status). Moreover, every participant had to evaluate the same painting (Portrait of Ambroise Vollard by Pablo Picasso 1910), so that answers to the survey are better interpretable compared to other researchers who were using blocks of art with many different styles, genres and artists (potential bias). This portrait is chosen after a small pre-test did validate my quest for a work of art made by a renowned artist, which is likeable and furthermore hard to ascribe to the artist due to the fact that the work is less famous. Twelve portraits, five sculptures and five drawings have been tested for their familiarity and

likability. The other works of art (sculpture and drawing) which made it to the questionnaire can be found in the appendix (1) at the end of this thesis.

As consumers with different levels of art expertise evaluate works of art in different ways (Leder et al, 2004), the main objective is gaining insights in whether an artists (both low and high status) can actually benefit from signaling different levels (single - vs multiple media) of spanning towards different kinds of potential consumers (expert & non-expert) in an experimental setting. The insights gained are both useful for academics as well as

practitioners due to the fact that there is not a lot of scientific knowledge yet concerning artists spanning multiple media and what the implications of their act could be. Moreover, so far I have not found any empirical evidence that such research (towards artistic media-spanning) exists.

Fascinated by the fact that some artist may actually benefit from signaling that they are capable of working in multiple media, others may not get this “benefit of the doubt”. This research is thus exploring what the consequences can be when acts of media-spanning are

7 communicated on consumers’ individual perceived financial market value, aesthetic

experience and furthermore the creativity perception of an artwork. Does it pay off for an artist to show that he or she is capable of working in multiple artistic media (generalist) or is it advisable to remain specialized in the eyes of a beholder (e.g. mere painter)? Implications are therefore useful for art galleries, artists themselves and even museums could benefit from the newly acquired knowledge. This could be either translated into higher sales, better museum experiences and even a more refined model of the aesthetic experience.

The structure of the thesis is as follows: it starts with a literature study in which different concepts are explained and defined and furthermore contains a critical discussion on the literature up to date concerning scholars from different fields of interest. This will lead up to the stated research hypotheses at the end of each paragraph. Moreover a survey-tool by Qualtrics is used in order to test participants. How this is done will be explained in the methodology section. After the methodology is explained, the results section will follow. Hereafter, you will find a comprehensive discussion based on existing theory and the newly gained knowledge from this study. A short conclusion based on the results will be provided at the final section.

8

Literature review

This part of the research critically discusses the literature on the influence of category-spanning activities, artistic status, art expertise, aesthetic experience and perceived financial value to date. At first, the literature on category-spanning will be examined. Furthermore, artists’ status will be discussed as potential mediator and finally, literature on aesthetic experience (A.E.), audience types and perceived financial value (P.F.V.) will be reviewed in order to complete this review.

Categories, genres and crossing them

The concept of categorical boundaries and the act of crossing them is being studied a lot by researchers now and in the past decades from within different fields of interest. For

example Fennel & Alexander conducted research on organizational boundary spanning (1987), Zuckerman on securities and the legitimacy discount (1999), Hsu et al. on movie box office revenues and eBay (2006), Kovács & Hannan on spanning and the consequences on online reviews of restaurants (2010), Negro & Leung on wineries who span different grape types (2013) and finally Durand & Paolella who focus on the categorization process and stretching of categories on organizational level (2012). For this research, the general

definition of a category by Negro & Leung will be used. They define a category as: “categories reflect shared understandings between market participants as to how similar producers and/or products can be clustered together and labeled (2012, P.685). But what happens when an artist does not only fit into one distinct category? This will make the particular person by definition a spanner, just like organizations who do not stick to their original organizational boundaries. Non media- spanning in the creative industries would thus imply that an artist being mere a painter (category label) is not working in other kinds of artistic media.

9 However, there are various examples of artists who do work in multiple media and

therefore should be assigned to more than one category. This makes the artist

interdisciplinary in terms of genres used (e.g. landscape and portrait) and/or on physical medium used (painting on canvas, bronze sculpture or pencil drawing). Even the

combination of different genres and media in one work of art can be understood as spanning (hybrid art form as explained in Levinson, 1984, P.6). In order to study category spanning, one needs to understand what these categories in terms of art can be. DiMaggio is the first researcher to state that it is impossible to divide the art world into discrete types and therefore needs to be seen as varying along several analytically distinct dimensions (1987). Art historians are furthermore keen on defining genres in terms of shared

conventions, but likewise do social relations enable them to identify producers by “schools” or “artistic movements” (DiMaggio, 1987, P.441). Although these art classification systems do exist and could be extremely useful in order to differentiate between – and categorize certain kinds of art, the main focus of this study is the act of using different physical media by an artist in its total oeuvre and the reaction of an audience towards this artist. ‘Media’ in this thesis is in line with the definition according to the Museum of Modern Art and is stated as follows: “The materials used to create a work of art, and the categorization of art based on the materials used (for example, painting [or more specifically, watercolor], drawing, sculpture”) (MoMA website, 2015).

As stated in the introduction, the implications of being a category-spanner can be

significant. Zuckerman for example, found that in financial markets, companies who straddle (thus being active in more than one category) established industry classes could face the problem of stock prices trading at a discount when the organization fails to get legitimacy by

10 specialized trade analysts. A producer (organization) tries to differentiate from other

competitors in order to be picked by its audience. However, when failing to meet audience criteria, it has been argued that an organization (and ditto stock price) is suffering due to a so called ‘illegitimacy discount’ (that is, the cost of illegitimacy) (1999). This implies that the audience couldn’t make sense of a company’s choice regarding specialization or

generalization and therefore penalizes the company in terms lower of stock prices. “Organizations or products are viewed as legitimate when they convey membership in a specific category, while those with attributes that span categories suffer from reduced attention, discounted evaluations, and poorer performances” (Zhao et al, 2013, P.4). Zuckerman furthermore argues that his theoretical scope is only valid in markets with two distinct characteristics. The first one is the fact that markets should have certain structural features consisting of: ‘a recognized set of product categories and an influential class of critics who specialize by category’. Additionally, he does argue that the perspective only applies to markets in which consumers have problems concerning product evaluation (Zuckerman, 1999). It should be said that both of this premises hold when studying the evaluation of artworks. Fine art artists do also work in an environment with recognized (mostly) sets of product categories and are being criticized by critics who often tend to specialize in certain categories (e.g. 19th century or Contemporary art) due to the fact that a lot of knowledge is required when professionally criticizing art. Furthermore does the art market consist of a lot of uncertainties that concern perceived quality. Therefore problems arise on aesthetic - and financial value evaluations of a work produced (Velthuis, 2003; Beckert & Rössel, 2004). Hsu et al. continue with and also argue the consequences of being a category-spanner by linking two approaches that used to be seen from only one side (2009). On the one hand they mention the producer-side view, and on the other an

11 audience-side view. “The producer-side view holds that spanning of categories reduces one’s ability to effectively target each category’s audience, which leads to decreasing appeal of audience members”, whereas according to the audience-side perspective, ‘’audience members refer to established categories to make sense of products’’. Products that incorporate features from multiple categories are perceived to be poor fits with category expectations and are therefore less appealing than category specialists” (Hsu et al., 2009, P. 150). The scholars focused their research on both E-bay auctions and U.S. feature-film projects (creative industries) in order to test their proposed hypotheses and suggest that both processes (audience- and producer-side) contribute to the penalties associated with category-spanning. These penalties could furthermore be more severe for low-status actors as researchers found while investigating the French cuisine business (Rao et al., 2005). The researchers tied given penalties (in terms of lower ‘critic’ ratings) to status and found that high-status actors are given more room for experimentation and therefore can more easily borrow from another category due to their unquestioned social acceptance, also known as status. On the other hand they argue that low-status actors do have less to lose when acts of spanning are used and this could therefore also be a viable strategy.

Other researchers like Kovács and Johnson also stress negative social consequences such as receiving less audience attention in their research next to the mere financial disadvantages (2013). Moreover, in another study, Kovács & Hannan (2010) put emphasis on the negative effects of being a category-spanner and summarize it into three bullet points. At first, there is a problem with learning and skills. As one is getting better in one category, they argue it’s really hard to gain the same skills and expertise in another category. Although they restrict their research to organizations, I argue that it is still useful for gaining insights into the

12 creative industries, and moreover gaining insights in individual artists. In their paper they refer to an earlier study of Hsu in which she calls organizations who span genres: ‘masters of none’, derived from the popular sentence: “Jacks of all trades and masters of none” (2006). This implies that organizations who work in multiple trades cannot be masters in all specific categories. Secondly, audiences tend to favor specialist instead of generalist. Consequently, generalists lose appeal towards their more specialist counterparts. Even when a company is able to be at the same level of expertise as a specialist in more than one category at the same time, they will still suffer difficulties convincing their targeted audience (P. 2). As a last point, Kovács and Hannan mention that people find it difficult to make sense of objects whose characteristics cause them to be assigned to multiple genres (Kovács & Hannan, 2010). An artist who is only linked to painting as medium could therefore be easier to make sense of than an artist spanning multiple media in their oeuvre.

Next to all negative consequences of being a category-spanner, there are also researchers stressing the benefits of being able to work in different categories at the same time and as already mentioned before, therefore able to attract larger audiences (e.g., Pontikes, 2012). Moreover, the merging of two different existing categories could have the same benefits by expanding market niches (Zhao et al., 2013 from Jensen, 2010; Karthikeyan & Wezel, 2010). Thus an artist working in multiple media is not merely seen as a painter, but could be seen as an artist who expands its market niches (innovative). Hsu et al for example argue that the film-industry is an example of an industry with a winner-takes-all structure. Hereby setting yourself apart from the rest could give you the advantage of exceptional attention gained by the strategy of spanning genres. Via the recombination of existing market categories, you will therefore be able to innovate. Moreover, the theory on recombination of categories in

13 order to create something totally different is already conceived by Schumpeter at the end of the 30’s (Hsu et al, 2012). Although both sides (positive and negative) of spanning are encountered in recent (and former) studies, this research is arguing in with line with the (most likely) negative effects of being a multi-media spanner and therefore the first hypotheses are stated as follows:

H1a: The perceived financial market value of artworks made by artists who predominately

work in one medium will be evaluated higher than those artists that span multiple media.

H1b: The perceived aesthetic experience of artworks made by artists who predominately

work in one medium will be evaluated higher than those artists that span multiple media.

These hypotheses imply that both the financial market value and aesthetic experience assigned by consumers to a work of art made by an artist working in multi-media are

expected to be lower than when an artist is only assigned to a single media (mere) painting. These hypotheses are furthermore in line with the reasoning of Sgourev and Althuizen who in both of their papers already found evidence for this towards artistic

14

The role of status as quality cue for the appreciation of art

As Sgourev and Althuizen already found in their research, artistic status is a significant contributor towards ones’ conception of artistic quality (Sgourev & Althuizen, 2015). Moreover, consumers tend to give high status artist more degrees of freedom for

experimentation in terms of categories spanned (Rao et al., 2005). Furthermore do many other researchers also focus on status as being an important determinant for one’s perception of quality. For instance, Waguespack & Sorenson found that high status film distributors get more lenient classifications by certifiers than do independent distributors. This gives them more space to play with certain categorical boundaries whereas low(er) status companies cannot profit from this advantage (2011). These implications in terms of status tend to be significant and therefore this research also puts emphasis on the status cues of artists. As with many of the definitions used in this thesis, ‘status’ of an artist is once again dubious. This is due to the fact that it is already difficult to unambiguous define the word: ‘artist’. When researchers are performing SA studies (status-of-the-artists), they tend to use the definition used by UNESCO (1980) when talking about artists (self-assessment criteria). This is done because when you are empirically investigating artists, it is fairly ambitious to state who’s an artist and who’s not, therefore using these self-assessment criteria as solution. This is summarized in the following sentence:

“Artist is taken to mean any person who creates or gives creative expression to, or re-creates works of art, who considers his artistic creation to be an essential part of his life, who contributes in this way to the development of art and culture and who is or asks to be recognized as an artist, whether or not he is bound by any relations of employment or association.” (Karttunen, 1998, P.7)

15 Describing artistic status is even more dubious, because you will have to know in whose eyes status is legitimate. Bain for example argues that “professional status is largely derived from the construction and maintenance of an artistic identity and its effective

communication towards others” (P.25). This implies that although someone is really skillful in painting, his or her professional status could still remain latent due to the fact that its existence is not communicated towards different beholders. Communication is therefore a leading factor in order to be recognized and the maintenance of this status is necessary to not be forgotten. Moreover does Bain discuss myths and dominant cultural stereotypes as concepts so that artists can “exaggerate and exploit their individuality and to feed into popular myths to reinforce their occupational authenticity” (Bain, 2005, P.29).

Althuizen and Sgourev also put emphasis on status as being an important cue in order to evaluate paintings more positively and this also applies to painters who cannibalize what they call: consistency and congruency in style (Althuizen & Sgourev, 2014). All this

knowledge on artistic status leads towards the following two research hypotheses:

H2a: The perceived financial market value of artworks made by artists will be evaluated

higher for those artists that are recognized as being high status compared to those who are not.

H2b: The perceived aesthetic experience of artworks made by artists will be evaluated

higher for those artists that are recognized as being high status compared to those who are not.

16

H3a: The perceived financial market value of artworks made by artists who predominately

work in multiple media will be evaluated higher for those artists that are recognized as being high status compared to those who are not.

H3b: The perceived aesthetic experience of artworks made by artists who predominately

work in multiple media will be evaluated higher for those artists that are recognized as being high status compared to those who are not.

The stated hypotheses are twofold. On the one hand you have got the fact that high status artists will be evaluated higher than low/no status artists. This is under no-spanning

conditions (mere painter – H2a – H2b). On the other hand, it is argued that when acts of media spanning are used, it is still preferred to have a high artistic status compared to no-status and low-no-status cues (H3a – H3b). This is due to the fact that no-status could lower the penalties associated with spanning. Therefore it is expected that that in any case, high status actors will have better scores on both aesthetic experience and perceived financial value even when media-spanning is at stake.

Aesthetic experience

The study of aesthetics has a long history which goes back to the time of the early Greeks (Madsen et al., 1993). When someone is being exposed to a work of art, one can undergo a so called ‘aesthetic experience’. According to older literature, aesthetic experience is divided into aesthetic perception (which is about objective quantities like: recognition and discrimination of a work of art) and aesthetic reaction. The reaction part is said to be subjective by nature, and so on, cannot be taught, controlled or tested (Madsen et al., 1993). Compared to today’s literature (e.g. Silvia, 2005; Pelowski & Akiba, 2010), early

17 literature on aesthetics experience was less of a process (more static) than it is being

thought of now (dynamic). For example, Leder et al. (2006) state that aesthetic appreciation of artworks involves an ongoing evaluation of the painting, which generates an incomplete impression, leaving room for further interpretation. Both emotions (affective) and thoughts (cognitive) are gathered into their model, which consists of the processing stages in one’s aesthetic experiences. Leder et al. (2006) found that for example higher understanding (as a cognitive process) of the artists’ intention and liking of the artwork tend to be leading factors for a more positive aesthetic experience. Other factors which they used to capture aesthetic experience are whether an observer found personal meaning, other elicited thoughts and emotional reactions due to exposure towards a work of art. Aesthetic value is furthermore influenced by context. Researchers argue that behavioral studies “have shown that presenting works artworks accompanied with titles, text, and other forms of cognitive information can significantly influence an observer’s reported evaluation of an artwork” (Kirk et al., 2009, P. 1125). Therefore this research is also linking positive contexts (high status / well known) towards higher values of aesthetic experience as can be seen in the hypotheses stated above (H2b & H3b).

The role of audience type on art valuations: art expertise

“Expertise is generally thought of as the possession of superior skills or knowledge in a particular area of study” (Herling, 2000, P.9). This can be anything between either job

related - or for example sport related knowledge. According to literature, art expertise could be held responsible for different effects on art appreciation like: preferences of for example high-brow art instead of low-brow art and better understanding of artistic intentions

(Augustin & Leder, 2006). This expertise used to be gathered via for example either studying and/or working in the arts sector, but due to the rise of internet, people are increasingly

18 becoming ‘self-made’ experts without having a certain degree. This convergence of

expertise levels even leads to tension between the amateurs and the experts and becomes particularly pronounced as popular meets high culture (Arora & Vermeylen, 2012; Sgourev & Althuizen, 2014). Arora & Vermeylen even wonder what the role of the traditional expert is going to be as non-experts get more and more involved in the art sector. Although some argue that the gap between experts and non-experts will dissolve, others are still

emphasizing that the role of the expert is indisputable in terms of their function as gatekeeper for the signaling of 'true' quality (Ginsburgh, 2003). Quality therefore can be interpreted as validated by experts via standardized measurement sets learned at for example art schools. However, this statement is also dubious as many others argue that quality is assessed via the eyes of the beholder (so anyone who views a certain piece of art). Modern - and contemporary art pave the path for almost endless innovation in terms of mixing styles, genres and material into one piece of art and so on, expertise tends to be needed in order to understand differences in the applied artistic strategy (Leder et al, 2006). In their second research on category-spanning (in-/congruency), Sgourev and Althuizen also made a distinction between levels of art observers’ expertise. Although they made use of (business) students as participants, they controlled for self-declared art knowledge, number of museum- and gallery visits and taught art lessons in order to divide the two groups of interest. According to their results, they argue to have found substantial differences in terms of valuation in the sense that congruency weighs more heavily for core consumers (‘expert’) than for casual consumers (‘non-expert’), regardless of the status of the artist (Althuizen & Sgourev, 2014). They argue that this is based on the fact that experts have a more refined body of domain knowledge and therefore stick to conservative stylistic norms, whereas non-expert consumers do not have these ‘boundaries’. Although non-experts are

19 also taught to be able to identify what they find aesthetically appealing and are furthermore able to give a monetary estimate towards a work of art, it is still argued that expert art observers will give higher (monetary) estimates. This is argued due to the fact that they are potentially more aware of what art can be worth in general. Hereby I am using ones market price estimate (fictive auction) instead of ones ‘willingness to pay’ which will be elaborated on in the next paragraph.

H4a: The perceived financial market value of artworks made by artists who predominantly

work in multiple media will be evaluated higher for those who are evaluated by expert consumers compared to those consumers who are less expertized.

H4b: The perceived aesthetic experience of artworks made by artists who predominantly

work in multiple media will be evaluated higher for those who are evaluated by expert consumers compared to those consumers who are less expertized.

Perceived financial value in monetary terms

“The concept of perceived value implies an interaction between a consumer and a product, is complex and multi-dimensional in nature, the value is relative by virtue of its comparative, personal and situational nature and furthermore is its value preferential, perceptual and cognitive-affective in nature” (Sánchez-Fernández & Iniesta-Bonillo, 2007, P.427). This interaction is based on the fact that a consumer is trying to find out what a product or service is worth to him or her and tries to catch the outcome mostly as a monetary measure. Your willingness to pay could correspond with your perceived value. However, there are also scholars arguing the scientific validity of this measurement due to a discrepancy between what one is willing to pay and what one is willing to accept (e.g.

20 Hanemann, 1991 & Coursey et al., 1987). They argue that their willingness to pay (W.T.P.) is mostly overrated compared to their willingness to accept. This could make the use of ones W.T.P. in a research setting as indicator for perceived value, less powerful. Thus, a different way literature tries to cope with perceived (financial) value is the approach of asking how much someone thinks something is worth when sold at auction (e.g. Leder et al, 2004; Sgourev & Althuizen, 2014). This approach is fairly interesting due to the fact that it’s not about what you want to pay for it, but is rather an approach of the conceived market value of a product. This is a better approach because even when a participant doesn’t for example like the subject of the painting, he is urged to give an estimate based on what others think instead of merely the individual. Therefore, perceived monetary value is used in this thesis to express the worth of art according to an individual’s estimate of an auction price (free market). It is known that this is not a representative measurement of the true value of the particular work of art which is mostly based on demand and supply plus an estimate of an expert valuator.

21

Methodology

This section of the thesis describes how the research is conducted. At first, the research design will be elaborated. The second part will describe the design of the experiment, its participants and furthermore the different experimental treatments which are used in the survey in order to study the consequences of media-spanning and artistic status on one’s financial- and aesthetic value estimate.

Research design

The purpose of this thesis is to find whether differences exist between two different groups (expert & non-experts art consumers) and their reaction towards an artist that spans and not spans different artistic media combined with both low/no and high artistic status cues. Their evaluation is measured by an estimate of their perceived financial market value for the painting and furthermore a measurement of their perceived aesthetic experience

(consisting of 4 items). In this way, the research is more or less an explanatory study (we already know from other study what could be potential influencers). Since the purpose of this thesis is to find whether experts and non-expert art observers are evaluating a painting differently for an artist that spans or not spans multiple media categories and who differs in terms of status, aesthetic experience and perceived financial value become the dependent variables, and the communicated ‘levels’ of spanning & status become the independent variables. In order to test the established hypotheses (differences between groups) given in the literature review, a survey in an experimental setting is used that will provide data for further analysis via IBMs’ SPSS. Although different platforms for the survey could be used, this research chooses to use Qualtrics due to its extensive possibilities for conducting research and the easy set-up for distribution. As stated in the introduction, only one

22 painting (the same for everyone: Ambroise Vollard by Pablo Picasso, 1910) had to be

evaluated by all participants, but the painting was accompanied with different contexts. The different contexts are therefore also the different research conditions to see whether they have implications on both groups’ aesthetic experience and perceived financial market value. These contexts can be found in the appendix (1) at the end of this thesis.

Respondents

290 participants volunteered to take part in the online survey, which did not require any art experience. However, only 231 respondents of them fully finished the survey (79%). Within those 231 respondents, no missing data were gathered due to the fact that answering all questions was obligatory (100%). Furthermore did they receive the link to the survey either found on Facebook (groups and friends), email (family/friends) or via other social media platforms (e.g. LinkedIn groups and Twitter). Although this method is said to potentially produce a selection bias, different arguments validate my choice and outweigh this bias argument. At first, gathering 200 + respondents in a short amount of time produces time constrains when you also want some more expertized art observers in your sample.

Therefore, large groups of known and unknown potential respondents – that turned out to differ in demographics- are approached in once in order to successfully meet the number of respondent’s criteria. This is moreover cost effective compared to other methods of

respondent collection (Handcock & Gile, 2011). Also, many other students and other researchers are using social media nowadays because of the potential reach of hyperlink to their surveys. In spite of this fact, the same strategy is used to gather my respondents. The strategy used can approximately be explained as snowball – and self-selection sampling (thus predominantly non-probability sample) due to the fact that there is interaction on different social media. When your questionnaire gets referred due to its content by initial

23 subjects to generate additional subjects, one can argue that it meets the criteria for calling it snowball sampling. Furthermore, snowball sampling is a data collection method used when participants are ‘hidden’ in terms of for example their specialized knowledge of something or in terms of demographics (Browne, 2005). Thereafter, recent research validates the use of social media networks (e.g. Facebook) in order to find these hard-to-reach respondents. They argue that the virtual response rate is higher than with traditional snowball sampling, which results in an increased sample size and therefore its representativeness, and they furthermore argue that the online questionnaire administration allows the quality of the information to be controlled and therefore avoids duplication of cases (Baltar & Brunet, 2012). Finally, it is important to know that when prospective respondents entered the survey, they were all randomly assigned to one of the seven experimental treatments (therefore, N = 200 + respondents were necessary). This is further elaborated in the method section below.

Demographics

Due to the fact that this research used self-selection and snowball sampling as data collection methods, it is known that the sample could not be a good reflection of a whole population. Therefore, caution has to be taken into account when generalizing data.

However, there are more than thirty respondents assigned per individual research condition so that the central limit theorem could hold (Field, 2009 P. 42). While 206 out of the 231 respondents are from Holland, it can be argued that the sample in general is Dutch tinted. Belgium, the United States and Germany complete this top 3 with 6, 4 & 4 respondents respectively. With respect to age -divided into 8 groups in this dataset- (< 16, 16-19, 20-24, 25-34, 35-44, 45-54, 55-64 and 65 or over), the most amount of respondents belong into the 25-34 group (33.8%). Furthermore, the least amount of people belong to the <16 (0.4%),

16-24 19 (4.3%) and the 65 or over category (4.8%). The difference between male and female participants is relatively small in favor of male respondents (122 vs. 109). Other

demographic statistics in this research concern education, employment and income. Only 2 respondents indicated to have “less than high school”, whereas the majority did a 4-years College degree (HBO in Dutch N=38). Continuing with the second highest group which is dictated by respondents with a master’s degree (N=51). It should therefore be argued that many people who volunteered to take part in this survey are highly educated compared to the Dutch average. In terms of employment, 74 of the 231 respondents were still students, but the biggest group is “employed for wages” with N = 85. Moreover, there are also 31 respondents self-employed. Nearly 50 % of the respondents earns between 0 and 25.000 euro. Only 8 out of 231 said to earn more than 100.000 euro on annual basis (gross income). Furthermore, 31 respondents state that they do not want to answer in which category they belong. Concluding we can state that this sample is not representative for a general

population since some demographic statistics are overpopulated compared to an average population and therefore caution needs to be taken into account when generalizing results. However, as the aim of this thesis is to differentiate between two distinct groups of art experts and non-experts it is still argued that the implications found can be useful due to sufficient dispersion of respondents.

Method

This quantitative research makes use of a 2 (media-spanning versus not-spanning) x 2 (low status vs high status) x 2 (naïve versus expert) between-subjects design. All participants were asked to evaluate one painting by Pablo Picasso (Portrait of Ambroise Vollard, 1910) accompanied with different signals (status / spanning). The first block is used as control (no status & no spanning cues), block two: no spanning – high status; 3: no spanning: low status;

25 4: spanning – high status; 5: spanning – low status; 6: no spanning – no status & 7: spanning – no status). The first section of the survey out of two consists of questions concerning participants’ gender, age, nationality, education, employment, income scale, the number of combined museum and art gallery visits last year and in their lifetime. Furthermore, two questions, one on interest in the visual arts and one concerned with the self-reported level of art knowledge complete the first part of the survey (personal information). Hereafter, participants had to look at the first block of works of art and answer questions based on familiarity of the painting (bias control), aesthetic experience (d.v, max score of 12: 4 x 3), creativity (control), and perceived financial market value (d.v.).

Experimental treatments

Status. Some participants were shown a signaling text consisting of either: “The artworks were made by a 20th _{century visual artist” (low status) or: “These artworks were made by} arguably one of the most popular and influential 20th _{century visual artist” (high status)} followed by the kind of medium from which the artwork is made (either oil on canvas, bronze sculpture, drawing or acrylic, charcoal and pastel on linen). As indicated by a small pretest, the work of art which needed to be evaluated was difficult to ascribe to a certain artist. I selected Pablo Picasso as the only artist for this research, because he is a renowned artist in different media but also made so many works of art that only a minor selection of them are world famous. While using the same artists, benefits arise due to out ruling artistic skill as explanation for data differences. Furthermore, Pablo Picasso is used by many

different scholars due to its’ extensive oeuvre, renowned status and quality in multiple disciplines (e.g. Miller, 2004, Sgourev & Althuizen, 2014). As every participant in the end is evaluating the same painting, no differences exist in terms of actual quality between the evaluated paintings and therefore this set-up minimalizes data noise.

26 Media-spanning. The media-spanning manipulation is based on the cues that are given to participants in terms of artistic media used. I constructed one set of only oil paintings (three in total) within the same genre (portrait) and style (cubist) for the artist working

predominately in a single media. The other block (again three works) consists of the same painting of Ambroise Vollard by Picasso, including a bronze sculpture and a work of art which is made with charcoal on paper; again made in the ‘same’ style and genre as the painting. Once more, this set-up is chosen in order to reduce biases concerning stylistic differences and thus it is expected that people will answer more accordingly than when mixing divergent artistic styles. This is for example explained in Sgourev & Althuizen (2014)

Control. One block is furthermore the control group due to the fact that its mere the painting of Ambroise Vollard that needs to be evaluated without any status or multiple media spanning signals (cues). Also, two additional control blocks are added in which the first one is mere having a single media artist (no spanning) without any status signals. The second one consists of a block with multiple media (spanning), but again without any status cues. These are used to test whether people react merely on spanning, left alone status.

Summarizing, every artwork is made by Pablo Picasso in the same style in order to minimize differences in likeability so that there is the least amount of noise (bias) within and between the research conditions. Furthermore, the seven blocks (visualized) can be found in the appendix at the end of this thesis.

27 Table 1: Overview of treatments and their cues

Research condition No spanning Spanning No status Low status High status No cues

1: Control x 2: NSp – HSt x x 3: NSp – LSt x x 4: SP – HSt x x 5: SP – LSt x x 6: NSp - NSt x x 7: SP - NSt x x Measures

Dependent variables. Leder et al. (2006) and Millis (2001) used a 7-point scale (strongly disagree – strongly agree) in order to measure subjects aesthetic experience (6 items) per painting. This research makes use of their validated measurement scale of aesthetic

experience, but in a slightly adjusted manner in order to decrease participants time needed to complete the survey (average time needed in total: 3-4 min.). Participants were asked four questions (three scale points: disagree (1), neutral (2), agree(3)) to indicate whether they liked the painting, whether they found the painting aesthetically appealing, whether the painting affected them emotionally and whether the painting evoked any thoughts in them. These items concern both cognitive as well as affective aspects of aesthetic

processing and are therefore useful to capture ones overall aesthetic experience (Leder et al, 2006). Excluded from the former method to measure aesthetic experience used by Leder et al. are understanding of the artists’ intention and personal meaning. Furthermore,

following Sgourev & Althuizen (2014), Getzels & Csikszentmihalyi (1969) and Ashenfelter & Graddy (2003), participants were then asked whether they could indicate the level of creativity felt on a single-item 10-point scale (low: 1 – high: 10). They emphasize that this question asks for a subjective, personal judgment of these paintings (Sgourev & Althuizen, 2014) and furthermore could serve as mediator between a participant and its estimates. The next question is concerned with the perceived financial market value and is stated as

28 follows: “What is the estimated amount of money (in €) the painting would probably receive when sold at an auction? (whole number, no commas needed). This is intended to be an ‘open’ question due to the fact that the choice for brackets with a specific width could steer participants into a certain (wrong) direction which could have the consequence of a biased estimate. It is known that this could generate non-normal data at the end.

Working variables. The number of museum and gallery visits last year and in their lifetime was measured on a five-point scale with 1 being “0 times” towards 5 being “more than 10 times” and 5 being “more than 15 times” when asked about lifetime museum visits. The next question is also in line with Sgourev & Althuizen (2014) and measures the self-declared level of art knowledge on a single-item consisting of five scales (1: low towards 5: high). Furthermore, participants were then asked to indicate their level of interest in the visual arts on a 5-point Likert scale (also 1: low and 5: high). These four questions in total are captured into one expertise variable (sum) in order to discriminate between non-expert and expert art consumers. In table 2, it is shown that the Cronbach’s Alpha for the 4 items in total show an internal reliability of 0.819. Working with a C.A. > 0.7 is for example recommended by Andy Field (Field, 2009)

Twenty is therefore the potential maximum and four is the minimum level of respondents’ art expertise. A median split like researchers Althuizen & Sgourev have used was also put to practice to divide the respondents in two groups based on expertise (M = 13, SD = 3.8) (2014).

Table 2: Reliability Statistics: Expertise Cronbach's

Alpha

N of Items

29 Control. The control variables consist of gender (male/female), age category (in brackets starting at 16 or lower, 16-19 etc. towards 65+), country of residence (dropdown list worldwide), education (less than high school towards doctoral degree), employment / studying (dropdown list) and gross income (brackets from 0-24.999 euro, 25.000-49.999 etc. towards 100.000 euro +). Participants also did have the possibility not to answer this

question by selecting: “don’t want to answer”. This suggestion is given out of respect for the respondents and because of the fact that some people are rather overwhelmed when asked to reveal their level of income.

30

Analyses and results

The next section is devoted to the results that were acquired by analyzing the data with SPSS. First of all, some descriptive statistics are given on familiarity, financial value and aesthetic experience. Due to the fact that normal distribution is not achieved in every single variable, other tactics are used in order to perform statistical analyses that do not follow the assumption of normal distributed data. These are called non-parametric tests as explained in Discovering SPSS (Field, 2009, chapter 15).

Descriptive statistics

For this research, it is important that people do not easily recognize Picasso as being the artist due to the fact that status cues given are important for the experimental treatments. Too easily recognizing the artist would thus mean that the status cues become less

important or useful and therefore potentially bias the outcomes. However, familiarity data suggest that from 231 participants, only 24 people “have seen this painting before (7)” compared to 168 participants who filled in “have not seen this painting before (1)”.

Furthermore, it can be seen in the table below that familiarity is fairly even distributed per condition. Other answers towards familiarity range between 2 and 6 indicate not knowing exactly whether they have seen this particular painting before.

Table 3: Participant assigned to research condition * Familiarity Cross tabulation 1.00 2.00 3.00 4.00 5.00 6.00 7.00

Participants’ research condition

1 Control 27 1 0 0 0 3 3 34

2: No Span - High Stat 25 1 1 1 1 2 3 34 3: No Span - Low stat 21 3 1 3 1 1 4 34 4: Span - High stat 24 2 0 1 0 1 4 32 5: Span - Low Stat 25 2 1 2 0 0 3 33 6: No Span - No Stat 24 3 1 1 0 0 4 33 7: Span - No stat 22 1 1 1 2 1 3 31

31 Familiarity is furthermore divided in a dichotomous variable with scores from 1 to 4 counted as 0 (not seen before) and 5, 6 and 7 as 1 (seen before). After looking for a frequency table, the

familiarity scores still tended to be evenly distributed among groups and therefore are not left out of the research. This is important because it is taught that familiarity could have some major

implications towards the height of outcome variables.

Financial value

Table 4 shows the descriptive statistics for all different treatments together in terms of perceived financial value. 231 respondents divided in seven research groups placed

estimates ranging between 10 euro to 50.000.000 euro, whereas the ‘lowest minimum’ is 10 euro in block 3 and block 2 is accountable for the ‘highest min. estimate’, namely: 500 euro. The average of all (P.F.V.’s) is 2.596.298,2 euro with the median being 30.000 euro. The results show that participants in block 3 have placed the highest mean estimate, namely: 4.1 million compared to 0.5 million in block 7 being the lowest. Overall it can be argued that the std. deviations are fairly extreme. However, this is not surprising with individual estimates which range between 10 euro and 50 million euro. Moreover, the average standard deviation (in €) is 8.3 million, whereas the highest standard deviation could be assigned to block 2: 12.0 million and the lowest SD could be found in block 7, counting 1.2 million euro.

Table 4: Descriptive Statistics of research condition with respect to perceived financial value (in Euro) N Minimum Maximum Mean: statistic Std. error Mean Std. Deviation

F.V. block 1 (Control) 34 350.0 20000000.0 1831341.2 855399.3 4987791.9 F.V. block 2 34 500.0 50000000.0 3869369.1 2051729.5 11963535.9 F.V. block 3 34 10.0 50000000.0 4109717.9 1917418.4 11180374.6 F.V. block 4 32 100.0 30000000.0 2319807.8 1212726.6 6860217.6 F.V. block 5 33 50.0 40000000.0 1976186.2 1257984.24 7226569.2 F.V. block 6 33 300.0 50000000.0 3369618.2 1606576.0 9229076.8 F.V. block 7 31 33.0 5000000.0 501449.5 208890.8 1163054.9

32 Below you can find the histogram, boxplots and normality distribution before data

transformation.

As expected before, the use of a perceived financial value estimate with no restrictions (e.g. brackets) will give a lot of outliers. This is because you are asking many different participants to evaluate a certain painting (see the graphs above) with most of them not knowing what it could be possibly worth. Moreover, the normality tests of the perceived financial value are for all research conditions significant, implying that normality of data distribution cannot be assumed, D (between 31-34) = 0.00, p < .05. These statistics can be found by using the Kolmogorov-Smirnov (K-S) test as shown in table 5 on the following page.

33

Transforming the data; financial value

While having non-normal distributed data, it can be a viable strategy to transform the perceived financial value into a logarithmic scale in order to get more normal looking data. After data transformation, the Kolmogorov-Smirnov (K-S) test reveals that for all the

research conditions, normality of data can be assumed. Furthermore, the skewness has also decreased to more conventional values. More in-depth: the first four blocks D (32 & 34) = 0.200, P > .05, are approximately normal. Block 5 however, shows a distribution close to the critical point D (33) = 0.059, P < .05, whereas block 6 and 7 are normally distributed again with D(33) = 0.159, P >.05 and D (31) = 0.200, P > .05 thus being significantly normal. With respect to skewness and kurtosis, normal distributions would have values around zero or close to (Field, 2009). As you can see in the table below, financial value is heavily (positive) skewed and suffers heavy kurtosis. Due to transformation of the financial value variable, the skewness and kurtosis levels are less critical than they were before.

Table 5: Tests of Normality for financial value and logarithmic financial value

Condition: Kolmogorov-Smirnov a

Shapiro-Wilk

Statistic df Sig. Statistic df Sig. Skew Kurtosis

Financial Value 1 0.390 34 0.000 0.416 34 0.000 3.224 9.790 2 0.418 34 0.000 0.356 34 0.000 3.673 12.793 3 0.433 34 0.000 0.428 34 0.000 3.150 9.712 4 0.417 32 0.000 0.387 32 0.000 3.497 11.775 5 0.433 33 0.000 0.303 33 0.000 4.935 25.770 6 0.359 33 0.000 0.413 33 0.000 4.364 21.418 7 0.376 31 0.000 0.491 31 0.000 3.062 9.265 Total 4.377 20.003 Logarithmic scale F.V. 1 0.100 34 .200* 0.960 34 0.240 0.418 -0.572 2 0.111 34 .200* 0.965 34 0.337 0.334 -0.332 3 0.113 34 .200* 0.968 34 0.401 0.125 -0.192 4 0.128 32 .200* 0.947 32 0.120 0.479 -0.768 5 0.149 33 0.059 0.964 33 0.341 0.510 -0.350 6 0.131 33 0.159 0.939 33 0.063 0.386 -0.994 7 0.115 31 .200* 0.977 31 0.738 -0.184 -0.484 Total 0.269 -0.548

*. This is a lower bound of the true significance.

34 Also detection of outliers has been tested. Although extreme values were found via data exploration, Tukey (1977) and later, Hoaglin & Iglewicz - & Tukey (1986 & 1987) found that you have to multiply (Q3 – Q1) with 1.5 (g-factor) in order to detect whether extreme values are possible outliers (Q1 minus number found and Q3 plus number found). However, in their next paper they concluded to have found too many outliers being detected that weren’t outliers after all. Thus they revised their so called g-factor towards 2.2. Due to the fact that (Q3-Q1) x 1.5 was already giving no outliers based on this logarithmic scale, 2.2 as factor would even expand the possible extreme numbers. Therefore, no outliers as such are detected for the logarithmic scale of perceived financial value. This is all visualized in the boxplot diagram on the right.

Aesthetic experience

Most of the total scores for aesthetic experience per research condition were neither normal distributed with significance levels D (31 – 34) = between 0.00 and 0.04, p < .05. The only exception in this trend is research condition 4, which has normally distributed data as indicated by D (32) = 0.144, p > .05. Again, the data for aesthetic experience show relatively high skewness, but lower kurtosis. This is also in line with the visual observations (e.g. boxplots on the next page). What is even more important is that respondents were predominantly positive about the work of art in terms of their aesthetic experience.

35

Table 6: Normality test for Aesthetic Experience

Even though different strategies are used and analyzed to enhance the normal distribution of the aesthetic experience data, none of the strategies did raise K-S scores significantly (higher P is more normally distributed). Moreover, logarithmic and square root

transformations are used without any additional benefits. These transformed data also tended to show significant K-S numbers, implying that data normality of data distribution is not achieved. Therefore, it has been chosen to use non-parametric tests that test data without the assumptions normally associated with parametric tests (e.g. normal distribution). Furthermore, detection of

outliers has been tested via the same manner described in the section above (financial value). Again, no outliers were found and no real life arguments were found that give us permission to exclude these extreme values from the dataset.

Participants' research treatment

Kolmogorov-Smirnova Shapiro-Wilk

Statistic df Sig. Statistic df Sig. Skew Kurtosis

AestheticEx 1 0.185 34 0.005 0.929 34 0.029 -0.701 -0.131 2 0.252 34 0.000 0.838 34 0.000 -1.329 1.415 3 0.154 34 0.040 0.920 34 0.016 -0.669 0.584 4 0.135 32 0.144 0.941 32 0.081 -0.509 0.460 5 0.190 33 0.004 0.940 33 0.066 -0.603 0.098 6 0.278 33 0.000 0.883 33 0.002 -0.844 0.092 7 0.235 31 0.000 0.836 31 0.000 -1.491 2.261 Total -0.837 0.170

36

Other variables

Table 7: Descriptive Statistics of creativity, familiarity & expertise

N Minimum Maximum Mean S.d. Skewness & s.d. Kurtosis & s.d.

Creativity 231 1.00 10.00 7.2857 1.63806 -1.479 .160 3.728 .319 Familiarity 231 1.00 7.00 2.0823 2.05515 1.662 .160 1.105 .319 Expertise 231 4.00 20.00 13.0173 3.78207 -.241 .160 -.560 .319

All respondents together generally turned out to find the painting by Pablo Picasso creative (mean = 7.29 out of 10). This is valid for all research conditions together (spanning + status). However, when only block 1 (no cues at all, thus control) is being analyzed, the creativity perception drops to M = 7,09. As already concluded above, the familiarity of the specific painting overall is very low, which is good for the validity of the results. This is also valid when only analyzing the control block which has a creativity mean value of 2. Furthermore, the median level of participants’ expertise based on a 4-item scale turned out to be 13. As explained before, thirteen is used as the splitting point which divides expert and non-expert art observers in the tests whereby 13 and under got labeled ‘non-experts’ and above 13 counts as ‘expert’. This has led to the following total distribution of participants.

As can be seen in the table 8 above, there are almost as many female as male participants. However, it should be said that almost double the non-expert art observers are male which means that in this dataset, female expert art observers are more present than male.

Table 8: Crosstab of gender * expertise

Non-Expert Expert Total

Gender Male: 0 79 43 122

Female: 1 42 67 109

37

Correlations

This section is devoted to the most important correlations between variables and starts with the correlation coefficient matrix which can be found in the appendix. Due to the fact that seven research conditions are tested, the most important significant correlations per group are given. This gives a better overview of possible relationships (direction and strength) between variables inside the groups. Furthermore, it is important to know that Spearman’s correlation rank order is used in contrast to Pearson’s correlation coefficient, due to the fact that most data is non-normally distributed.

Block 1: Control

Familiarity correlates positively with both art expertise (r = 0.490, p < 0,01) and financial estimate (r = 0.494, p < 0.01), implying that when you recognize the painting more, you will be more likely to have higher levels of art expertise and a higher financial estimate. When one recognizes a painting by Picasso without any status cues, it is most likely that he is more expertized and therefore knows a little bit more about the possible valuation of millions for a certain work. Furthermore, creativity and aesthetic experience and creativity and financial value are also positively correlated, meaning that higher levels of creativity lead to higher levels of aesthetic experience (r = 0.468, p < 0.01) and higher levels of financial estimates (r = 0.450, p < 0.01). Aesthetic experience also shows a moderate positive correlation with financial value (r = 0.468, p < 0.01). Respondents in this block with higher levels of A.E. are thus more likely to give higher financial estimates.

Block 2: No spanning – High status

In block two, a significant positive correlation between gender and expertise is found (r = 0.478, p < 0.01). This means that when one is getting a ‘higher gender’ (0 = male, 1 = female)

38 she is also more likely to be more expertized in the arts. Familiarity and expertise are again positive correlated (r = 0.461, p < 0.01), but familiarity furthermore correlates with creativity (r = 0.455, p < 0.01). Creativity and aesthetic experience also show a moderately positive significant correlation (r = 0.528, p < 0.01) however, creativity does furthermore positively correlate with financial value (r = 0.464, p < 0.01). Thus people who perceive the painting to be creative, tend to give higher financial – and aesthetic estimates. Art expertise also shows a significant positive correlation with financial value (r = 0.461, p < 0.01). Therefore,

respondents with higher levels of art expertise tend to give the painting higher financial estimates.

Block 3: No spanning – low status

Again, familiarity does show positive correlation with financial estimate (r = 0.556, p < 0.01). Moreover, also creativity does correlate positively with both aesthetic experience (r = 0.481, p < 0.01) and perceived financial value (r = 0.456, p < 0.01). Just like in block 1, aesthetic experience shows a significant positive correlation towards perceived financial value (r = 0.439, p < 0.01).

Block 4: Spanning – high status

Familiarity with the painting does show significant positive correlation with creativity (p < 0.05), expertise (p < 0.01), aesthetic experience (p < 0.05) and financial estimate (p < 0.01). Recognizing Picasso as the artist will thus significantly improve many variables. Furthermore, creativity does show a positive correlation with both aesthetic experience (r = 0.587, p < 0.01) and perceived financial value (r = 0.419, p < 0.05). In this block, expertise tend to positively correlate with aesthetic experience (r = 0.407, p < 0.05). Moreover, gender positively correlates with aesthetic experience (r = 0.502, p < 0.01) and expertise (r = 0.545, p < 0.01).

39 Block 5: Spanning – low status

Again, familiarity could be held accountable for positive correlation with creativity (p < 0.01), expertise (p < 0.01), and perceived financial value (p < 0.01). As in most other research conditions, creativity and aesthetic experience show a positive correlation (r = 0.547, p < 0.01), but now creativity also correlates with perceived financial value (r = 0.515, p < 0.01). Furthermore, art expertise positively correlates with financial value (r = 0.374, p < 0.05).

Block 6: No spanning – no status

In this block, familiarity correlates positively with three other variables, namely: creativity (r = 0.405, p < 0.05), expertise (r = 0.684, p < 0.01) and financial value (r = 0.584, p < 0.01). Creativity furthermore correlates with aesthetic experience (r = 0.361, p < 0.05). Expertise again does correlate positively with perceived financial value (r = 0.374, p < 0.05).

Block 7: Spanning – no status

Education is negatively correlated to creativity perception (r = -0.399, p < 0.05), implying that the more educated you are (scale is ascending), the lower the perception of creativity for this particular work of art. Familiarity on the other hand is again significant positively correlated with creativity (r = 0.424, p < 0.05), expertise (r = 0.410, p < 0.05) and perceived financial value (r = 0.437, p < 0.05).

Summarizing, we can conclude that familiarity with the work of art positively correlates with most other variables. Also creativity could be held accountable for these interactions. However, the degree to which they do this and whether this effect is significant for explaining the leading dependent variables for this thesis will be explained via regression analysis later.

40

Testing hypotheses with non-parametric t-tests.

In this section the 4 x 2 hypotheses (a/b) are tested. In order to test for significant

differences, mean score measures are used to differentiate between the perceived financial and aesthetic scores versus the mean estimates of other groups. This with respect to

different levels of art expertise, status and spanning (between research groups). While normality of the data could not be assumed (see appendix), non-parametric tests are used. According to Field, these tests are also known as assumption-free tests, because they make fewer assumptions about the type of data on which they can be used (2009, P. 540). Most of the tests work by ranking the data on the basis of finding the lowest score (1: minimum) and so on until the highest score is found (x: maximum). This leads to high scores that are

represented by large ranks and low scores that are represented by low ranks. The analysis is thereafter based on the ranks instead of the actual data. Important to know is that the data sample is large enough so that for most tests the ‘asymptotic method’ can be used in

contrast to ‘exact tests’ when the sample of interest turned out to be smaller (last statistical test).

Mann-Whitney U and effect size (r)

Before answering the research hypotheses, the general groups are tested for differences between aesthetic experience and perceived financial value. Scores that are found differ on the basis of high status vs low / no status to see whether status cues did change the

perception in general. Therefore research condition 2 and 4 are combined and set up against 3,5,6,7. In the tables below you can find the differences between having a high status compared to no / low status. The results indicate that the mean rank for A.E., creativity and F.V. differences are not significant (P = 0.976, 0.681 & 0.681). We can therefore conclude that status is not a very strong cue in this thesis.